Sleep Stage Assessment in Very and Extremely Preterm Infants: Exploring the Relationship Between Behavioural Classification and Quantitative EEG Features
Summary
Identification of sleep and wake states is important for the clinical and neurophysiological assessment of infants born preterm admitted to the neonatal intensive care unit (NICU). As the importance of different sleep stages on neurodevelopment and long-term prognosis is currently unravelling, there is a high demand for quantification of sleep stages. Electroencephalography (EEG), especially two-channel amplitude-integrated EEG, is a commonly used neuromonitoring tool in daily care of preterm infants due to its ease of use. The present study aims to investigate whether quantitative EEG (qEEG) features can differentiate the different stages of sleep in very and extremely preterm infants.
Three-hour behavioural sleep observations were performed for 17 very and extremely preterm infants who were born before 30 weeks of gestation, within the first three days of life. The behavioural sleep stage classification scores were acquired using an observational score validated for preterm infants <30 weeks postmenstrual age. Several qEEG features were extracted from raw signals of the two-channel (a)EEG, including burst features (proportion spontaneous activity transients, SAT%; inter-SAT percentage, ISP; and inter-SAT interval, ISI), interhemispheric synchrony (Activation Synchrony Index, ASI) and absolute and relative spectral power of the delta frequency band. Differences of these qEEG features among different sleep stages were analysed by ANOVA.
Significant differences were found in several qEEG features at different sleep stages, mostly between Active Sleep (AS) and Quiet Sleep (QS). Specifically, the SAT%, ASI and absolute delta power of AS were significantly higher than that of QS, while ISP and ISI were higher for QS than for AS.
Several qEEG features were identified that can differentiate different sleep stages, and thus, could be beneficial to the improvement of sleep stage classification. The present study sets the foundation for the development of an automatic sleep assessment tool using EEG for the very to extremely preterm population. Ultimately, this will eventually lead to the improvement of neurodevelopmental outcome in the very to extremely preterm infants in NICUs worldwide.